Cervical Disc Arthroplasty A Technology Overview

Transcription

1 Cervical Disc Arthroplasty A Technology Overview ADOPTED BY THE AMERICAN ACADEMY OF ORTHOPAEDIC SURGEONS BOARD OF DIRECTORS March 8, 2010 This Technology Overview was prepared using systematic review methodology, and summarizes the findings of studies published as of September 9, on cervical disc arthroplasty. As a summary, this document does not make recommendations for or against the use of cervical disc arthroplasty and it should not be construed as an official position of the American Academy of Orthopaedic Surgeons. Readers are encouraged to consider the information presented in this document and reach their own conclusions about cervical disc arthroplasty. The Academy has developed and is providing this Technology Overview as an educational tool. Patient care and treatment should always be based on a clinician s independent medical judgment given the individual clinical circumstances. 1 v

2 Disclaimer This technology overview was developed by an AAOS physician volunteer task force based on a systematic review of the current scientific and clinical information and accepted approaches to treatment and/or diagnosis. This technology overview is not intended to be a fixed protocol, as some patients may require more or less treatment or different means of diagnosis. Clinical patients may not necessarily be the same as those found in a clinical trial. Patient care and treatment should always be based on a clinician s independent medical judgment, given the individual patient s clinical circumstances. Disclosure Requirement In accordance with AAOS policy, all individuals whose names appear as authors or contributors to this technology overview filed a disclosure statement as part of the submission process. All panel members provided full disclosure of potential conflicts of interest prior to developing the key questions contained within this technology overview. Funding Source This technology overview was funded exclusively by the American Academy of Orthopaedic Surgeons who received no funding from outside commercial sources to support the development of this document. FDA Clearance Some drugs or medical devices referenced or described in this technology overview may not have been cleared by the Food and Drug Administration (FDA) or may have been cleared for a specific use only. The FDA has stated that it is the responsibility of the physician to determine the FDA clearance status of each drug or device he or she wishes to use in clinical practice. Copyright All rights reserved. No part of this technology overview may be reproduced, stored in a retrieval system, or transmitted, in any form, or by any means, electronic, mechanical, photocopying, recording, or otherwise, without prior written permission from the AAOS. Published 2010 by the American Academy of Orthopaedic Surgeons 6300 North River Road Rosemont, IL First Edition Copyright 2010 by the American Academy of Orthopaedic Surgeons 2 v

3 SUMMARY OF RESULTS Summaries of the data pertaining to the four key questions addressed in this Technology Overview are provided. All four questions and the studies included to address each question compare the outcomes of patients treated with cervical disc arthroplasty (CDA) to patients treated with anterior cervical disc fusion (). QUESTION #1: What patient characteristics predict successful outcomes in patients who undergo cervical disc arthroplasty (CDA) compared to patients who undergo anterior cervical discectomy and fusion ()? The outcomes of interest for this question included the following: previous surgeries per patient, all demographics available, age, sex, smoking status, workmen s compensation status, narcotic use, opioid use, analgesic use, use of TENS Unit, and any ongoing pain management if evaluated in a study. Most studies considered for this question, did not report or conduct the appropriate statistical analyses such as regression or multiple regression to examine predictive patient characteristics with patients considered to have successful clinical patient-oriented outcomes. At 24 months, the authors of one Level study with 147 patients reported no statistically significant difference in the percentage of successful patients treated with CDA compared to successful patients treated with in regards to the continuation of the use of strong narcotics and muscle relaxants (See Table 7). These results are inconclusive as to what patient characteristics predict successful outcomes in patients treated with cervical disc arthroplasty compared to patients treated with anterior cervical disc fusion. QUESTION #2: Do patients with herniated cervical disc who present with arm pain with or without neck pain and are treated with a cervical disc arthroplasty(cda) have equal or better clinical outcomes than patients treated with anterior cervical discectomy and fusion ()? Five level studies were considered to address this question. Below, we present a brief summary of the results of the outcomes addressed in the studies considered for this overview. Please see pages for further information of the following outcomes: Neck Disability Index scores Three of the four studies we included reported that at earlier follow-up durations (1.5-3 months) patients treated with CDA had statistically significantly lower NDI scores than patients treated with fusion. Results at longer follow-up durations are inconclusive. Neck Disability Index success rate 3 v

4 Two of the three Level studies we included reported that, at 3 months, patients treated with CDA had statistically significantly higher NDI success rates. No statistically significant differences were reported at later follow-up durations. Neurologic success rate Two of the three Level studies we included reported that at all follow-up durations there were no statistically significant differences between treatment groups. One Level study reported that patients treated with CDA had statistically significantly higher neurologic success rates at 12 months. Pain (VAS) o Neck Pain Four of the five Level studies we included reported no statistically significant differences in neck pain at earlier follow-up durations (1 6 months). One study reported patients treated with CDA had statistically significantly less neck pain than patients treated with. The results reported at later follow-up durations are inconclusive. o Arm Pain Three of the four Level studies we included reported no statistically significant differences in arm pain scores at all follow-up durations. One study reported that at 24 months, patients treated with CDA at multiple levels had statistically significantly less arm pain compared to patients treated with at multiple levels. Short-form-36 The results reported by three of the Level studies included for this overview are inconclusive. Return to Work Two Level studies we included reported no statistically significant differences in the number of patients who returned to work at 24 months and one study reported similar results for patients returning to heavy work at 24 months. 4 v

5 QUESTION #3 Do patients with herniated cervical disc who present with arm pain with or without neck pain and are treated with a cervical disc arthroplasty have equal or better revision rates, and/or complication rates than those treated with anterior cervical discectomy and fusion? Four level studies were included for this question. The results of secondary surgical procedures reported by three of the studies are inconclusive as the authors of these studies do not report or measure secondary surgical procedures of patients similarly, and therefore, the results cannot be compared. The results of any adverse events of patients reported by four of the studies considered for this question are also inconclusive. QUESTION #4 For patients, what is more economical, cervical disc arthroplasty or anterior cervical discectomy and fusion as defined by hospital (LOS) and length of time to return to work (RTW)? Four Level studies were included to address this question. No statistically significant differences were reported in the length of hospital stay for patients treated with CDA compared to patients treated with. Patients treated with CDA returned to work in statistically significantly fewer days (range days) than patients treated with. 5 v

6 INTRODUCTION The prospect of achieving relief of radicular arm and neck pain, while at the same time maintaining spine segmental motion and thus eliminating adjacent segment degeneration, is very appealing. With the advent of the artificial disc, this scenario may be possible. Increasing numbers of artificial discs are becoming available for use along with a growing body of data with longer follow-up. The question remains however; "Does the currently available evidence answer the question of whether artificial disc replacement is as good as or superior to anterior cervical fusion in relief of neck and arm pain when used to address similar clinical scenarios as anterior cervical fusion?" Evidence-based medicine utilizes a three legged stool approach to arrive upon appropriate clinical decisions. The best available evidence (1) is incorporated with the (2) physician s experience and (3) patient values to select the best available treatment recommendation for an individual patient. Often, in incorporating newer technology into clinical practice, physician experience is limited and exuberant marketing and unrealistic expectations of the value of a new technique can unduly influence patient values. Given the potential inherent weaknesses of two of three of the legs of the evidence-based medicine triad, the best available evidence becomes that much more important in clinical decision making early in the incorporation of new technology into a clinician s practice. The purpose of this technical review is to examine the best available evidence on cervical artificial disc replacement when compared to the current gold standard of anterior cervical fusion and plating. The AAOS presents this evidence using a process that includes a meticulous literature search combined with a methodical evaluation of relevant manuscripts to present to the reader the best available evidence. It is left to the reader to reach their own conclusions. METHODS OVERVIEW This report was developed using the methods of a systematic review 1. We began by having a panel of physicians frame four Key Questions, and next developed rules (inclusion criteria) for determining what information we would include (The full list of criteria appears in Appendix I); articles were only included if they met the a priori criterion. Finally, we conducted comprehensive literature searches (See Appendix ) to ensure that the data we considered are not biased in favor of any particular point of view. Thereafter, we evaluated the quality of the relevant studies (including their methods of analysis) considered and compared their results, and summarized this information. The program TechDig 2.0 (Ronald B. Jones, Mundelein, Illinois) was used to estimate means and variances from studies presenting data only in graphical form. Also, we calculated the variance of the arcsine difference to confirm statistical significance (p < 0.05) 2 and converted one-sided probability values to two-sided values in order to consistently compare the results of all the studies included in this overview. INCLUDED ARTICLES Our search identified 2054 citations that were potentially relevant to this overview and that could potentially meet our inclusion criteria. Of these, 7 studies 3-9 were included to address the key questions (See Table 1). Six of these studies compared the outcomes of 6 v

7 patients treated with single level cervical disc arthroplasty to patients treated with single level fusion with adjunctive augmentation. One study 6 compared the outcomes of patients treated with cervical disc arthroplasty (CDA) at multiple levels to patients treated with anterior cervical disc fusion () at multiple levels. The majority of the studies in this Overview 4, 6, 7, 8 (n= 566) included patients with a herniated cervical disc and/or cervical degenerative disc disease. These studies did not include patients with moderate or severe or marked spondylosis. Two studies 3, 9 (n=297) included patients with spondylosis and neck or arm pain (radicular) and or functional/neurological deficits [but excluded patients with severe spondylosis or ankylosing spondylitis (chronic spondylosis)]. One study 5, does not report whether or not patients with spondylosis were included. Table 1. Included studies and corresponding questions Author Q1 Q2 Q3 Q4 Heller, - - Murrey Cheng, Mummaneni, Nabhan, Anderson, Riew, study was included for (x) question - study not included for (x) question DEVICE PROPERTIES The studies included in this overview reported that patients were treated with a metal-onpolymer or metal-on-metal artificial cervical disc; patients treated with disectomy and fusion received an anterior cervical plate with varying adjunct augmentation and with or without a cage ( See Table 2). Information regarding the size of the artificial discs used to treat patients in these studies was not reported in detail. Any information the authors reported regarding disc size was given in general terms. Specifically, the authors reported the sizes available for a specific type of the artificial disc but did not disclose the number of patients who received any given size. Nor was disc size compiled for specific groups of patients identified; therefore, disc size information was not useful for this overview. 7 v

8 Table 2. Device properties for patients treated with arthroplasty vs. fusion. Author CDA Heller, metal on polyurethane plate w/ bone allograft Anderson, 2008 metal on polyurethane plate w/ structural allograft Riew, 2008 metal on polyurethane plate w/ allograft Cheng, 2008 metal on polyurethane plate w/ iliac crest autograft Mummaneni, et al 2007 metal on metal plate w/ cortical ring allograft Murrey, metal on polyethylene plate w/ bone allograft Nabhan, 2007 metal on polyethylene plate w/ cage a a bone graft not specified DEVICE RECALL INFORMATION The U.S. Food and Drug Administration have issued recalls associated with two of the devices used to treat patients with CDA or reported by four of the studies included in this overview. Please see Table 3 and Table 4 for further information regarding the recalls for these devices. Table 3. FDA recall classifications Recall class Recall Description Recall examples Class I Dangerous or defective products that predictably could cause serious health problems or death Food containing botulinum toxin, food with undeclared allergens, a label mix-up on a lifesaving drug, or a defective artificial heart. Class Class I Products that might cause temporary health problem, or pose only a slight threat of a serious nature Products that are unlikely to cause any adverse health reaction but that violate FDA labeling or manufacturing laws A drug that is under-strength but that is not used to treat life-threatening situations. A minor container defect and lack of English labeling in a retail food. 8 v

9 Table 4. CDA and device recall information Recall Recall Author Device Class Number Number Heller, I Z Mummaneni, 2007 Mummaneni, 2007 Mummaneni, 2007 CDA Z Z I Z Anderson I Z Riew I Z Reason for Recall Screwdriver handle breakage Step drills from Lot may have been mis-etched Implant mis-seating; Variance in size between the trial and the implant could cause the implant to be improperly seated. Screwdriver handle breakage Screwdriver handle breakage Screwdriver handle breakage Date Posted Dec. 28, 2005 Dec. 11, 2003 Dec. 19, 2007 Dec. 28, 2005 Dec. 28, 2005 Dec. 28, v

10 QUALITY OF THE LITERATURE The quality of evidence is an important and critical step in the systematic review process. In studies investigating the result of treatment, we assessed the quality of the evidence for each outcome at each time point reported in a study, not simply the overall quality of a study. Our approach follows the recommendations of the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) working group 10 as well as others. 11 We evaluated quality on a per outcome basis rather than a per study basis because quality is not necessarily the same for all outcomes and all follow-up times reported in a study. For example, a study might report results immediately after patients received a given treatment and after some period of time has passed. Often, nearly all enrolled patients contribute data at early follow-up times but, at much later follow-up times, only a few patients may contribute data. In this scenario, one would have more confidence in the earlier data than in the later data. The fact that we would assign a higher quality score to the earlier results reflects this difference in confidence. We assessed the quality of treatment studies using a two-step process. First, we assigned a level of evidence to all results reported in a study based solely on that study s design. Accordingly, all data presented in randomized controlled trials were initially categorized as Level I evidence, all results presented in non-randomized controlled trials and other prospective comparative studies were initially categorized as Level. We next assessed each outcome at each reported time point using a quality questionnaire and, when quality standards were not met, downgraded the level of evidence (for this outcome at this time point) by one level. OUTCOMES CONSIDERED We preferentially included patient-oriented outcomes over surrogate outcomes. This was partly because patient-oriented outcomes are important to patients and indicate, without the need for extrapolation, whether an intervention is effective. 12 Patient-oriented outcomes include pain, quality of life, ability to perform activities of daily living, and revision surgery. On the other hand, surrogate outcomes substitute for a clinical event of true importance. 12 Common surrogate outcomes include laboratory tests, biomarkers, range of motion, and radiographic findings. Unlike use of patient-oriented outcomes, use of surrogate outcomes can be misleading, and can even make harmful treatments look beneficial. 13 We found patient-oriented evidence for every question. MINIMAL CLINICALLY IMPORTANT IMPROVEMENT Wherever possible, we considered the effects of treatments in terms of the minimal clinically important improvement (MC) in addition to whether their effects were statistically significant. The MC is the smallest clinical change that is important to patients, and recognizes the fact that there are some treatment-induced statistically significant improvements that are too small to matter to patients. The values we used for MCs are derived from a published studies investigating the Visual Analogue Scale, and 14, 15 the Neck Disability Index. 10 v

11 Table 5 MC of outcomes Outcome Measure MC (points) Pain VAS (0-100) 15 NDI The associated descriptive terms in this technology overview and the conditions for using each of these terms, are outlined in the following table: Table 6 Descriptive terms for results with MC Descriptive Term Condition for Use Clinically Important Statistically significant and lower confidence limit > MC Possibly Clinically Important Not Clinically Important Negative Inconclusive Statistically significant and confidence intervals contain the MC Statistically significant and upper confidence limit < MC Not statistically significant and upper confidence limit < MC Not statistically significant but confidence intervals contain the MC POWER To assess the power of an outcome to detect a statistically significant difference we determined whether the number of patients in the study was sufficient to detect a small, medium, or large effect, while assuming an alpha of 0.05 as the significance level, 80% power, and Cohen s definitions of small, medium, and large effects (a small effect is d = 0.2, a medium effect is d = 0.5, and a large effect is d = 0.8). 16 When a study with a nonsignificant difference was unable to detect a medium or large effect it was categorized as low power. Studies able to detect medium effects or with statistically significant differences were categorized as high power. Six of the seven included studies for this Overview were categorized as having high power and one study 17 was categorized a low powered study v

12 QUESTION 1: What patient characteristics predict successful outcomes in patients who undergo cervical disc arthroplasty compared to patients who undergo anterior cervical discectomy and fusion? SUMMARY OF RESULTS The outcomes of interest for this question included the following: previous surgeries per patient, all demographics available, age, sex, smoking status, workmen s compensations status, narcotic use, opioid use, analgesic use, use of TENS Unit, and any ongoing pain management if evaluated in a study. Most studies considered for this question, did not report or conduct the appropriate statistical analyses such as regression or multiple regression to examine predictive patient characteristics with patients considered to have successful clinical patient-oriented outcomes. At 24 months, the authors of one Level study with 147 patients reported no statistically significant difference in the percentage of successful patients treated with CDA compared to successful patients treated with in regards to the continuation of the use of strong narcotics and muscle relaxants (See Table 7). These results are inconclusive about what patient characteristics predict successful outcomes in patients treated with cervical disc arthroplasty compared to patients treated with anterior cervical disc fusion. The results of the included study 3 that addressed this question (See Appendix V) reported unreliable Level evidence for the outcomes (See Appendix I). The results of the included study are unreliable because they were reported as a composite measure. Composite outcome measures, such as overall success as reported in this included study 3 are unreliable because each individual outcome might not equally influence or contribute to the overall significance of the estimated effects of the given treatment; hence, less important outcomes can be more influential than more serious outcomes (i.e. death and/or serious adverse events). Studies suggest examining the results of the individual outcome measures along with the results of the composite outcome measures to ensure a comprehensive examination of the effects of a given treatment Other studies considered for this question, did not report or conduct the appropriate statistical analyses or did not compare patients treated with CDA to patients treated with, therefore these studies were not included (See Appendix V). 12 v

13 STUDY RESULTS Table 7 Medication use of successful patients Treatment Author LOE N a, b Duration Group Medication use c, d p-value Study e AAOS f Murrey et I CDA % p = 0.1 p =.065 al. months I % a Number of patients considered an overall success b Overall success defined by the authors as the percentage of patients with Neck Disability Index success ( 15 pt improvement/ reduction from baseline), maintenance or improvement in neurologic status (measured by motor function, sensory function, and tendon function; all three conditions had to be satisfied in order to be considered a success), no serious implant related adverse event or adverse event related to the implant procedure, or secondary surgical procedure. c Medication use includes the use of strong narcotics and muscle relaxants. d Strong narcotics defined as schedule 2 drugs with high abuse and high dependency risk e The authors reported the results of one-tailed tests. We converted the values reported by the authors to two-tailed values. f test of arcsine difference 13 v

14 QUESTION #2: Do patients with herniated cervical disc who present with arm pain with or without neck pain and are treated with a cervical disc arthroplasty have equal or better clinical outcomes than patients treated with anterior cervical discectomy and fusion? SUMMARY OF RESULTS To address this question, we included five studies 3-7 that examined five outcomes (See quality Table 26-Table 30). The data were all Level except for three of the outcomes reported by Mummaneni The results reported by Mummaneni et al at 24 months were Level I data because patient follow-up at this duration was <80% (see section on quality of literature). Based on this flaw, the outcomes data reported at 24 months were not included to address this question; the NDI, neurologic success and VAS pain results at earlier follow-up durations were included to address this question. NECK DISABILITY INDEX (NDI) SCORES Four studies 3-6 reported NDI scores of patients treated with CDA compared to patients treated with at various follow-up durations (See Table 8 and Figure 1). One study 6 reported NDI results of patients treated with CDA at multiple levels compared to patients treated with at multiple levels. Patients with lower NDI scores are considered to have less disability when performing activities of daily living compared to patients with higher NDI scores. Three of the four studies reported that, at earlier follow-up durations (1.5 3 months), patients treated with CDA had statistically significantly lower NDI scores than patients treated with but the differences are not considered as clinically important. Results at longer follow-up durations are inconclusive in that one of the four studies reported statistically significant differences in favor of patients treated with CDA at 6 months and one study reported no statistically significant results; two of the four studies reported statistically significant results in favor of CDA at 12 months and at 24 months, two of the three studies reported statistically significant results in favor of patients treated with CDA (Table 8). NDI SUCCESS RATE Three studies 3-5 reported NDI success rates as the percentage of patients with 15 pt improvement/ reduction from baseline. Two studies 3, 5 reported that, at 3 months, patients treated with CDA had statistically significantly greater NDI success rates than patients treated with (See Table 9). Three studies reported no statistically significant differences in NDI success rates of patients at later follow-up durations (6 24 months). One study 4 reported that, at 24 months, the NDI success rates of patients treated with CDA was statistically significantly noninferior (margin of inferiority, δ = 0.10) to the success rate of patients treated with (See Table 9). NEUROLOGIC SUCCESS RATE Three studies 3-5 reported unreliable (See question #1 for section on composite outcomes) and inconclusive results of neurologic success rates defined as the maintenance or improvement in neurologic status from baseline measured by motor function, sensory function, and tendon function; all three conditions had to be satisfied in order for a patient 14 v

15 to be considered a success. Details were not reported regarding how motor, sensory, and tendon function were measured. All three studies reported no statistically significant differences in neurologic success rates at earlier follow-up durations (See Table 10). One study reported that at 12 months, patients treated with CDA had statistically significantly greater neurologic success rates than patients treated with (See Table 10). One study reported that the neurologic success rates of patients treated with CDA were statistically significantly noninferior to patients treated with at 24 months. NECK PAIN (VAS) The results reported by five Level studies 3-7 are inconclusive. One study 3 reported neck pain results that are incomparable to the results reported by the other studies included to address this question. Four of the five studies reported no statistically significant differences in the neck pain of patients at earlier follow-up durations (1 6 months); one study reported patients treated with CDA had statistically significantly less neck pain than patients treated with (See Table 11 - Table 12 and Figure 2). At later follow-up durations (12-36 months), two 4, 6 of the five studies reported patients treated with CDA had statistically significantly less neck pain than patients treated with but in one of these two studies, patients were treated with either CDA or at multiple levels of the cervical spine. One study 3 reported that, at 3 months, patients treated with CDA had statistically significantly less neck pain intensity than patients treated with ; no statistically significant differences in neck pain scores were reported by the authors of this study at all other follow-up durations (See Table 12). ARM PAIN (VAS) Five studies 3-7 reported the results of arm pain scores of patients up to 36 months following treatment (See Table 13 - Table 14 and Figure 3). Four studies reported no statistically significant differences in the arm pain scores of patients at 36 months. One study 6 reported patients treated with CDA at multiple levels had statistically significantly less arm pain than patients treated with at multiple levels. SHORT FORM-36 Three studies 4-6 reported the SF-36 physical component summary (PCS) scores and two studies reported the mental component summary (MCS) scores up to 24 months following treatment. One of the three studies 4 reported that patients treated with CDA had statistically significantly greater improvements in PCS and MCS scores up to twelve months following treatment but the difference was not statistically significant at twentyfour months (See Table 12 and Table 15). One of three studies 5 reported no statistically significant differences in PCS and MCS scores at all follow-up durations. Two studies 4, 5, categorized as having high power, report conflicting results at 6 months; one study reported statistically significant results where as the second study reported no statistically significant results. One study 6 reported that, at 12 and 24 months, patients 15 v

16 treated with CDA at multiple levels had statistically significantly higher PCS scores than patients treated with at multiple levels (See Table 15, Table 16,; Figure 4. Figure 5). RETURN TO WORK Two studies 3, 4 reported no statistically significant difference in the percentage of patients who returned to work at 24 months and one study 3 reported similar results for patients returning to heavy work (See Table 17). Please see question #4 for the results of patient s length of time to return to work. 16 v

17 STUDY RESULTS Table 8. Neck Disability Index scores a b Author Treatment N Baseline 1 week 1.5 months 3 months 6 months 12 months 18 months 24 months Heller, CDA (nr) nr 22.5 (nr)* 17.6 (nr) 16.1 (nr) 15.1 (nr) nr 16.2 (nr) c, f (nr) nr 31.2 (nr) 22.4 (nr) 21 (nr) 18.8 (nr) nr 19.2 (nr) Murrey, CDA (15.0) nr 29.1 (18.5) 21.7 (16.7) 23.0 (19.3) 22 (20.4) 21.2 (19) 21.4 (20.2) d (14.5) nr 30.7 (16.4) 25.9 (19.8) 22.2 (20.1) 21.7 (18.5) 22.2 (18.5) 20.5 (18.4) Mummaneni CDA (nr) nr 27.1 (nr) 20.7 (nr) 21.7 (nr) 20.6 (nr) nr - et al, c, f (nr) nr 32.1 (nr) 26.8 (nr) 24.5 (nr) 23.4 (nr) nr - Cheng, CDA (nr) 21.3 (nr) nr 19.7 (nr) 14.3 (nr) 12 (nr) nr 11 (nr) 2008 e (nr) 24.0 (nr) nr 20.2 (nr) 19.2 (nr) 18 (nr) nr 19 (nr) Abbreviation: CDA, cervical disc arthroplasty;, anterior cervical disc fusion; ANOVA, analysis of variance; ANCOVA, analysis of covariance. a Data are presented as mean and standard deviation (SD) unless otherwise indicated. b NDI range of scores is c ANCOVA, pre-op score used as covariate. d Wilcoxon rank-sum test e ANOVA f Heller et al and Mummaneni et al reported the results of one-tailed tests. For comparability, we converted the values reported by both authors to two-tailed values. * nr refers to not reported Authors report no statistical significant difference but do not provide p-value. Number of patients at baseline Values presented in bold italic are significantly greater than CDA; p v

18 NDI Score differences 2 (CDA Score) - ( Score) wk Heller Murrey Mummaneni Cheng -10 Time Post-surgery (Months) Figure 1. Difference between mean NDI scores for the CDA and groups over time. * A negative score indicates improved function in favor of the CDA group and positive score indicated improved function for the group. 18 v

19 Table 9 Neck Disability Index success rates N CDA Author Duration CDA: % of patients Mummaneni, 2007 Mummaneni, 2007 Murrey, Mummaneni, 2007 Mummaneni, months 3 months 3 months 6 months 12 months 531 (274:257) 498 (257:241) 202 (101:101) 492 (259:233) 493 (265:228) p-value a study b AAOS b 75.4% 68.40% p = p = % 73.80% p = 0.008* p 0.01* nr* nr p = 0.001* n/a 81.4% 77.20% p = 0.27 p = % 79.10% p =.215 p =.962 Heller, 24 months 423 (229:194) 86.0% 78.9% p = 0.07 p = Murrey, 24 months 191 (99:92) 79.80% 78.30% p = p = Heller, 423 d 24 months 86.0% 78.9% p = n/a (229:194) Abbreviation: CDA, cervical disc arthroplasty;, anterior cervical disc fusion. a The authors reported the results of one-tailed tests. For comparability, we converted the values reported by both authors to two-tailed values. b P-values are for Fisher exact test unless otherwise indicated. c P-values reported from test of arcsine difference. * nr refers to not reported. Values presented in bold italic are statistically significant; p d P-values reported from tests of noninferiority. Table 10 Neurological success rates Author LOE Duration N CDA: CDA p-value a % of patients study b AAOS c Heller, months 423 (229:194) nr* nr ns n/a± Mummaneni, months 498 (257:241) 92.0% 87.0% p = p =.086 Murrey, et al. 6 months 209 (103:106) 94.6% 85.1% p =.092 n/a Mummaneni, et al months 492 (259:233) 92.5% 90.0% p =.318 p = v

20 Author LOE Duration N CDA: CDA p-value a % of patients study b AAOS c Mummaneni, et al months 493 (265:228) 92.5% 85.0% p =.024 p =.009 Murrey 24 months 209 (103:106) 90.9% 88.0% p =.638 p =.404 Heller, 24 months 423 (229:194) 93.9% 90.2% p = p = Heller, d 24 months 423 (229:194) 93.9% 90.2% p< n/a Abbreviation: CDA, cervical disc arthroplasty;, anterior cervical disc fusion. a All three studies reported the results of one-tailed tests. For comparability, we converted the values reported by the authors to two-tailed values. b P-values are for the Fisher exact test unless otherwise indicated. c P-values reported from test of arcsine difference. d P-values reported from tests of noninferiority. Author LOE Duration N CDA CDA: Mean (SD) p-value d Table 11 Neck Pain (VAS) a, b, c Heller, 463 d, e Baseline (242:221) 75.4 (19.9) 74.8 (23.0) p =.765 Mummaneni, 2007 e Baseline Nabhan, et al 2007 f Baseline Cheng, 2008 f Baseline Nabhan, et al Immediate post-op Heller, d, e 1.5 months Mummaneni, 2007 e 1.5 months Heller, Mummaneni, 2007 Heller, 3 months 3 months 6 months 541 (276:265) 41 (20:21) 62 (31:34) 41 (20:21) 451 (237:214) 531 (274:257) 439 (234:205) 498 (257:241) 423 (227:196) 67.6 (nr) 68.6 (nr) nr 60 (12.0) 62 (9.0) p = (nr) 71 (nr) nr 35 (6.0) 29 (7.0) nr 32.7 (nr) 37.5 (nr) p = (nr) 19.9 (nr) p = (nr) 32.8 (nr) p = (nr) 17.6 (nr) p = (nr) 32.7 (nr) p < v

21 Author LOE Duration Mummaneni, 2007 Heller, Mummaneni, 2007 Nabhan, et al months 12 months 12 months 12 months Cheng, 2008 f 12 months Heller, 24 months Cheng, 2008 f 24 months N CDA: 492 (259:233) 431 (235:196) 493 (265:228) 40 (19:21) 62 (30:32) 424 (230:194) 62 (30:32) CDA Mean (SD) p-value d 17.6 (nr) 19 (nr) p = (nr) 28.1 (nr) p = (nr) 19.4 (nr) p = (2.0) 15 (3.0) nr 19 (nr) 25 (nr) nr 23 (nr) 30.3 (nr) p = (nr) 26 (nr) p = Nabhan, et al months 17 (4.0) 25 (4.0) p = (19:21) Abbreviation: CDA, cervical disc arthroplasty;, anterior cervical disc fusion; ANCOVA, analysis of covariance. a Data are presented as mean and standard deviation (SD) unless otherwise indicated. b Reported results calculated by multiplying the intensity score by the frequency score. c The range for neck pain scores is points. d Heller et al and Mummaneni et al reported the results of one-tailed tests. For comparability, we converted the values reported by both authors to two-tailed values. e ANCOVA, pre-op score used as covariate. f Values converted to scale for comparability. 21 v

22 Neck Pain score differences 4 2 (CDA Score) - ( Score) post-op Heller Mummaneni Nabhan Cheng Time Post-Surgery (months) Figure 2. Difference between mean neck pain scores for the CDA and groups over time. * A negative score indicates improved function in favor of the CDA group and positive score indicates improved function for the group. 22 v

23 Table 12. Neck pain intensity and frequency (VAS) a N CDA Author LOE Outcome Duration CDA: Mean (SD) Murrey, et al. Neck Pain - Intensity 1.5 months 209 (103:106) p-value b 30.5 (24.9) 25.6 (21.3) ns* Murrey, et al. Neck Pain - Frequency 1.5 months 209 (103:106) 37.9 (30.6) 33.1 (28.3) ns Murrey, et al. Neck Pain - Intensity 3 months 209 (103:106) 24.1 (24.1) 27.2 (24.6) p <.05* Murrey, et al. Neck Pain - Frequency 3 months 209 (103:106) 34.1 (33.8) 36.6 (33.6) ns Murrey, et al. Neck Pain - Intensity 6 months 209 (103:106) 27.4 (27.7) 27.4 (26.6) ns Murrey, et al. Neck Pain - Frequency 6 months 209 (103:106) 37.9 (36.6) 35.9 (33.8) ns Murrey et al. Neck Pain - Intensity 12 months 209 (103:106) 25.1 (28.7) 27.2 (26.6) ns Murrey et al. Neck Pain - Frequency 12 months 209 (103:106) 34.3 (36.9) 36.9 (34.3) ns Murrey, et al. Neck Pain - Intensity 18 months 209 (103:106) 25.1 (25.4) 25.9 (25.4) ns Murrey, et al. Neck Pain - Frequency 18 months 209 (103:106) 30.8 (33.6) 34.3 (33.1) ns Murrey, et al. Murrey, et al. Neck Pain - Intensity Neck Pain - Frequency 24 months 24 months 209 (103:106) 209 (103:106) Abbreviation: CDA, cervical disc arthroplasty;, anterior cervical disc fusion. a Data are presented as mean and standard deviation (SD) unless otherwise indicated. b Wilcoxon rank-sum test. * nr refers to not statistically significant; authors do not report p-value. Values presented in bold italic are significantly greater than CDA; p (28.2) 24.4 (26.6) ns 34.1 (35.9) 30.6 (33.6) ns 23 v

24 Table 13. Arm pain (VAS) a CDA Author LOE Duration N CDA: Mean (SD) p-value b Heller, Baseline 463 (242:221) 71.2 (19.5) 71.2 (25.1) p = Mummaneni, 2007 c Baseline 541 (276:265) 59 (nr) 62.9 (nr) nr Nabhan, et al Baseline 41 (20:21) 73 (14.0) 72 (15.0) P = 0.1 Cheng, 2008 Baseline 62 (30:32) 71 (nr) 72 (nr) nr Nabhan, et al Immediate post-op 41 (20:21) 18 (4.0) 16 (4.0) nr Heller, 1.5 months 451 (237:214) 19.5 (nr) 22.1 (nr) p = 0.3 Mummaneni, 2007 c 1.5 months 531 (274:257) 13.3 (nr) 13.3 (nr) p = 1.0 Heller, c 3 months 439 (234:205) 19.3 (nr) 19.9 p = Mummaneni, months 498 (257:241) 12 (nr) 12.4 (nr) p = Heller, 6 months 423 (227:196) 20.4 (nr) 22.5 (nr) p = Mummaneni, et al months 492 (259:233) 14.3 (nr) 13.3 (nr) p = 1.0 Heller, 12 months 431 (235:196) 16.5 (nr) 21.3 (nr) p = Mummaneni, months 493 (265:228) 14.8 (nr) 15.7 (nr) p = Nabhan, et al months 41 (20:21) 14 (2.0) 15 (3.0) p = 0.06 Cheng, months 62 (30:32) 18 (nr) 24 (nr) ns Heller, 24 months 424 (230:194) 19.1 (nr) 21.5 (nr) p = v

25 Author LOE Duration N CDA: CDA Mean (SD) p-value b Nabhan, et al months 41 (20:21) 12 (3.0) 19 (2.0) nr Cheng, months 62 (30:32) 14 (nr) 27 (nr) p = Nabhan, et al months 41 (20:21) 12 ( 3) 17 ( 2) p = 0.06 Abbreviation: CDA, cervical disc arthroplasty;, anterior cervical disc fusion a Data are presented as mean and standard deviation (SD) unless otherwise indicated b Wilcoxon rank-sum test c Heller et al and Mummaneni et al reported the results of one-tailed tests. For comparability, we converted the values reported by both authors to two-tailed values. * nr refers to not statistically significant; authors do not report p-value ns refers to not statistically significant; authors do not report p-value Values presented in bold italic are significantly greater than CDA; p 0.05 Arm Pain score differences 4 (CDA Scores) - ( Scores) post-op Time Post-surgery (months) Heller Mummaneni Nabhan Cheng Figure 3. Difference between mean arm pain scores for the CDA and groups over time. * A negative score indicates improved function in favor of the CDA group and positive score indicates improved function for the group. 25 v

26 Table 14. Arm pain intensity and frequency (VAS) Author LOE Outcome Duration Murrey, Murrey, Murrey, Murrey, Murrey, Murrey, Murrey, Murrey, Murrey, Murrey Murrey, Murrey, Arm Pain - Intensity Arm Pain - Frequency Arm Pain - Intensity Arm Pain - Frequency Arm Pain - Intensity Arm Pain - Frequency Arm Pain - Intensity Arm Pain - Frequency Arm Pain - Intensity Arm Pain - Frequency Arm Pain - Intensity Arm Pain - Frequency 1.5 months 1.5 months 3 months 3 months 6 months 6 months 12 months 12 months 18 months 18 months 24 months 24 months N CDA: 209 (103:106) 209 (103:106) 209 (103:106) 209 (103:106) 209 (103:106) 209 (103:106) 209 (103:106) 209 (103:106) 209 (103:106) 209 (103:106) 209 (103:106) 209 (103:106) CDA Mean (SD) Abbreviation: CDA, cervical disc arthroplasty;, anterior cervical disc fusion a Wilcoxon rank-sum test * ns refers to not significant ; authors do not report p-value p-value a 22.7 (28.3) (23.7) ns* 25.2 (31.6) 20.4 (28.3) ns 15.8 (22.7) 18.8 (27.5) ns 17.6 (27.0) 19.8 (28.8) ns 19 (28.5) 18.8 (22.5) ns 19.8 (30.3) 22.3 (29.8) ns 17.2 (26.5) 22.5 (30.1) ns 18.2 (30.1) 27.5 (36.9) ns 18.2 (25) 19.2 (24.5) ns 18.4 (26.0) 22.4 (29.3) ns 19.9 (26.5) 17.3 (23.5) ns 20.6 (27.5) 22.4 (31.3) ns 26 v

27 Table 15. SF-36 physical component scores a N CDA Author LOE Duration CDA: Mean (SD) Heller, Mummaneni, 2007 Cheng, 2008 Baseline Baseline Baseline Heller, b c 1.5 months Heller, Heller, 3 months 6 months Mummaneni, 2007 b c 6 months Heller, Mummaneni, 2007 Cheng, 2008 Heller, Cheng, months 12 months 12 months 24 months 24 months 463 (242:221) 541 (276:265) 62 (30:32) 451 (237:214) 439 (234:205) 423 (227:196) 492 (259:233) 431 (235:196) 493 (265:228) 62 (30:32) 424 (230:194) 62 (30:32) p-value 32.6 (6.7) 31.8 (7.2) p = (nr) * 32.2 (nr) nr 35 (nr) 34 (nr) nr 41.3 (nr) 38.2 (nr) p < (nr) 43.9 (nr) p = (nr) 45.1 (nr) p = (nr) 43.1 (nr) p = (nr) 45.5 (nr) p = (nr) 43.6 (nr) p = (nr) 46 (nr) p = (nr) 46.3 (nr) p = (nr) 45 (nr) p = Abbreviation: CDA, cervical disc arthroplasty;, anterior cervical disc fusion a Data are presented as mean and standard deviation (SD) unless otherwise indicated b Heller et al and Mummaneni et al reported the results of one-tailed tests. For comparability, we converted the values reported by both authors to two-tailed values. c Results based on ANCOVA; pre-op score used as covariate * nr refers to not reported Values presented in bold italic are statistically significant in favor of CDA; p v

28 SF-36 PCS score differences 6 (CDA Scores) - ( Scores) Heller Mummaneni Cheng Time Post-surgery (months) Figure 4. Difference between SF-36 scores for the CDA and groups over time. * A positive score indicates improved function in favor of the CDA group and negative score indicates improved function for the group. 28 v

29 Table 16. SF-36 mental component scores a N CDA Author LOE Duration CDA: Mean (SD) Heller, Mummaneni, et al 2007 Heller, Baseline Baseline 1.5 months 463 (242:221) 541 (237:257) 451 (237:214) p-value b, c 42.3 (12.5) 44.6 (11.6) p = (nr)* 42.8 (nr) nr 51.4 (nr) 48.5 (nr) p <0.002 Heller, 3 months 439 (234:205) 52.6 (nr) 50.8 (nr) p = Heller, 6 months 423 (227:196) 53 (nr) 50.8 (nr) p <0.002 Mummaneni, et al months 492 (259:233) 49.3 (nr) 49.5 (nr) p = 1.0 Heller, 12 months 431 (235:196) 52.5 (nr) 51.6 (nr) p = Mummaneni, months 493 (265:228) 50.6 (nr) 49.2 (nr) p =.105 Heller, 24 months 424 (230:194) 51.7 (nr) 51.7 (nr) p = 0.54 Abbreviation: CDA, cervical disc arthroplasty;, anterior cervical disc fusion; ANCOVA, analysis of covariance. a Data are presented as mean and standard deviation (SD) unless otherwise indicated b Heller et al and Mummaneni et al reported the results of one-tailed tests. For comparability, we converted the values reported by both authors to two-tailed values. c Results based on ANCOVA; pre-op score used as covariate * nr refers to not reported Values presented in bold italic are statistically significant in favor of CDA; p v

30 SF-36 MCS score differences 4 (CDA Score) - ( Score) Heller Mummaneni -3 Time Post-surgery (months) Figure 5. Difference between SF-36 scores for the CDA and groups over time. * A positive score indicates improved function in favor of the CDA group and negative score indicates improved function for the group. Table 17. Percentage of patients who returned to work N CDA Author LOE Outcome Duration CDA: % Heller, Murrey, et al. Murrey, et al. Return to work Return to work Return to heavy work 24 months 24 months 24 months 301 (157:144) 175 (87:88) 115 (54:61) Abbreviation: CDA, cervical disc arthroplasty;, anterior cervical disc fusion a p-value reported from test of arcsine difference p- value a 76.8% 73.6% p = % 80.0% p = % 44.7% p = v

31 QUESTION #3 Do patients with herniated cervical disc who present with arm pain with or without neck pain and are treated with a cervical disc arthroplasty have equal or better revision rates, and/or complication rates than those treated with anterior cervical discectomy and fusion? SUMMARY OF RESULTS To address this question, we included four Level studies 3, 5, 6, 8 that reported secondary surgical procedures, adverse events and complications of patients treated with CDA at a single level or at a single level and one study 6 reported the complications of patients treated with either CDA or at multiple levels.(see quality Table 31 and Table 32). One study 3 reported unreliable results due to the use of a composite measure (See question #1 for discussion of composite outcomes). SECONDARY SURGICAL PROCEDURES Three Level studies reported inconclusive and incomparable results of secondary surgical procedures of patients treated with CDA or at 24 months. Secondary surgical procedures included revisions, supplemental fixation, implant removal and reoperations (See Table 18). One of the three studies 3 included reported unreliable results of the device success rate of patients treated with CDA compared to patients treated with (See Table 18) (See question #1 for explanation of the reliability of composite measures). One 8 of the three studies reported no statistically significant differences in the overall reoperation rates of patients treated with CDA compared to patients treated with (See Table 18). This study also reported that statistically significantly fewer patients treated with CDA required reoperations at any level of the cervical spine. The author reported that the difference was statistically significant but AAOS calculations cannot confirm this (See Table 18). One 6 of the three studies reported no secondary surgical procedures occurred in patients treated with CDA at multiple levels compared to patients treated with at multiple levels. ADVERSE EVENTS The results reported by four Level studies 3, 5, 6, 8 regarding the number of adverse events of patients treated with CDA compared to the adverse events of patients treated with are inconclusive. One study 8 excluded complications or any adverse events not meaningful to the treatment and that had no affect on the results of patients (i.e. post-op facelift surgery or being hit with a golf ball). Two studies reported the severity of adverse events based on the World Health Organization (WHO) severity scale. See Table 19 for information and description of each grade. One 3 of the four studies reported that at 24 months, there is no statistically significant difference in the number of adverse events that occurred in patients treated with CDA compared to patients treated with (See Table 20). One 5 of the four studies reported, that at 36 months, patients treated with CDA had statistically significantly fewer 31 v

32 serious adverse events than patients treated with but AAOS calculations cannot confirm this (See Table 20). One study 8 reported that within the peri-operative period, patients treated with CDA had statistically significantly more surgical related adverse events or acute neurologic adverse events (Grades 1-4). The authors report the difference as statistically significant, but AAOS calculations cannot confirm this (See Table 21 and Table 22). One 5 of the four studies reported that, within the peri-operative period, no statistically significant differences in the number of patients with adverse events (See Table 21). One 6 of the four studies reported that one patient treated with CDA at multiple levels had a deep vein thrombosis and one patient treated with at multiple levels had dysphagia. 32 v

33 STUDY RESULTS Table 18. Device success and the percentage of patients with secondary surgical procedures at 24 months Author LOE N CDA: Outcome CDA % % p-value Study AAOS a Murrey et al. 209 (106:103) Device Success b 98.1% 91.5% p = 0.06 c p = Anderson, (242:221) Reoperation (cervical) d 5.4% 7.7% p = p = Anderson, (242:221) Reoperation e (thoracolumbar) 7.0% 8.1% p = 0.56 p = Anderson, (242:221) Reoperation (Total ) 7.0% 8.1% p = 0.15 p =0.649 Cheng, et al (30:32) Revisions 0% 0% n/a* n/a Abbreviation: CDA, cervical disc arthroplasty;, anterior cervical disc fusion. a p-value reported from test of arcsine difference. b The percentage of patients who did not require reoperation, revision, supplemental fixation or implant removal. c Murrey et al reported the results of one-tailed tests. For comparability, we converted the values reported by the authors to two-tailed values. d Reoperation at the index, adjacent, or both levels of the cervical spine. e Reoperation in the upper extremity, shoulder, carpal tunnel, ulnar nerve transposition, thoracic outlet release. Values presented in bold italic are statistically significant; p n/a refers to not applicable. 33 v